BestDose : switch AUC calculation to log-linear trapezoidal

[R-garreau]

Switch AUC calculation from linear to log-linear

Linear :

$$\sum_{n=1}^k { \frac{ C_n + C_{n+1} }{2} \times (t_{n+1} - t_n) \qquad (1)} $$

For log linear :

• if $$C_{n+1} > C_n$$ : then linear formula (1)
• if $$C_{n+1} > C_n$$ : then use (2)

$$\sum_{n=1}^k { \frac{ C_{n+1} - C_n }{log \left( \frac{C_{n+1}}{C_n} \right)} \times (t_{n+1} - t_n ) \qquad (2)}$$

Principle

[mhovd]

Do you want log-linear also for the absorption phase?

[R-garreau]

then it would be a log-log AUC.
I think if you time step is small enough it won't matter anyways but you computation time may suffer.

Maybe there would be a way to automate that :

1. firt performing linear regression between C0 (t=0) and Cmax (Tmax)
2. if R² < 0.8 (or any other value that we may chose) then chose log-log AUC
3. if R² > 0.8 stay with log linear